--On 12. Juli 2010 14:39:19 -0400 Tom Lane <tgl@sss.pgh.pa.us> wrote:
> Yeah. Past discussions of this have leaned to the viewpoint that we
> should disallow the above, ie it should not be possible to have NOT NULL
> on the parent unless all the children have it too. Nobody's got round
> to implementing it though. There's a TODO item for it: see
> "Move NOT NULL constraint information to pg_constraint"
FYI, i'm currently working on it, see
<https://commitfest.postgresql.org/action/patch_view?id=312>
Bernd